A subset of the famous Google "Quick Draw" dataset ( https://github.com/googlecreativelab/quickdraw-dataset )
The original dataset is very noisy and big, this dataset was created for images similarity algorithms or for evaluation of these.
- classes num: 182
- images num: 18.200 (100 images per class)
- images: 128x128 / RGB
- the data was downloaded and sorted by similarity with the "Shape Context" algorithm, and then checked manually (yes... manually π)
- too similar and too noisy classes are filtered and deleted (this is the reason why it contains only 182 classes, and not 345)
The answer is yes.
This is only one example of noise removed in this subset.